It is an interesting idea, but I'm not sure of your premise that questions with more UVs are harder to answer.
The odd tide of votes come and goes, I sure don't know why some things get UVed so much and not others.
I can for sure say that UVs on a question have no correlation to the difficulty of the question.
Also UVs on questions should not indicate the popularity of a question; views do that. Now it is true that when a question is popular, it gets a lot of views and thus more UVs. But UVs are a measure of how many people found that post useful, not "liked" as in facebook popular.
Take this question as an example; it is about a little known feature, and has only gotten 650 views in 2 years (not a lot). Surely you can't say that it accumulated many UVs because of its popularity, and it sure is not a difficult thing to fix.
In fact some of my most UVed answers are answering simple questions.
How to convert a single vertex into a plane within a mesh?
How to Model a Keyboard key?
Is there a way to make comments in the node editor?
See Why is voting important? and What is voting up?; they make it clear that when you vote it represents much more then just, for example "you liked the picture," but rather you are playing a part in the whole site wide sorting of content best to worst.
However, after saying all that, if you could find a metric to find the truly difficult questions and award good answers a bonus I would be for that.
It might be something like unanswered for more then 3 weeks with more then 4 UVs on the question and less then 100 views, then the bonus kicks in.